Software as a Medical Device (SaMD)
Russia's framework for SaMD
Roszdravnadzor published methodological guidelines in 2018 establishing when standalone software qualifies as a medical device and how it should be registered. Russia was one of the earlier regulators to issue specific SaMD guidance.
The guidelines establish:
- Criteria for classifying software as a medical device based on its intended use
- Principles for classification — which risk class (I–III) applies to different software types
- Criteria for non-compliance that may be identified during the FGBU expert review
- Assessment algorithms for technical documentation review
- A list of applicable national standards for software evaluation
When software is a medical device
Software is treated as a medical device in Russia when the manufacturer declares an intended use that falls within the scope of Article 38 of Federal Law 323-FZ — for example:
- Software that analyses medical images to assist in diagnosis
- Clinical decision support software that makes patient-specific treatment recommendations
- Software monitoring vital signs and generating alerts
- Telemedicine platforms providing clinical diagnosis or treatment recommendations for individual patients
General-purpose analytics tools, electronic health records without decision-support functions, and administrative software are generally not considered medical devices.
Classification of SaMD
Software classified as a medical device receives a risk class (I, IIa, IIb, or III) based on the same criteria applied to hardware devices. The potential harm from software failure or error is the key criterion. AI/ML software used in high-stakes clinical decisions (oncology diagnosis, cardiac rhythm analysis) will typically be Class IIb or III.
OKPD-2 codes for software
The official OKPD-2 classification list was updated to include codes specifically for medical software. The correct OKPD-2 code must be assigned when preparing the registration dossier.
Technical documentation for SaMD
The technical dossier for a software medical device must include:
- Software description and architecture
- Intended purpose and intended user
- Version control and lifecycle management documentation
- Cybersecurity risk assessment
- Clinical and analytical validation data
- Instructions for use in Russian
AI and machine learning
AI/ML-based medical devices follow the same SaMD framework. Russia has engaged with IMDRF work on AI in healthcare (Russia chaired IMDRF during the 2019 meeting where AI was a key topic). There is no separate AI-specific regulation, but the FGBU review process for AI devices examines validation methodology carefully.
Roszdravnadzor methodological guidelines on medical software (2018): roszdravnadzor.gov.ru